ask Categorical for ranking function

Author: jeet63

pandas/core/algorithms.py

@@ -988,8 +988,8 @@ def _get_data_algo(values, func_map):
     elif is_unsigned_integer_dtype(values):
         f = func_map['uint64']
         values = _ensure_uint64(values)
-    elif is_categorical(values) and values._ordered:
-        f = func_map['float64']
+    elif is_categorical_dtype(values):
+        f = values._funcs_for_rank()
         values = values._values_for_rank()
     else:
         values = _ensure_object(values)

pandas/core/categorical.py

@@ -1404,6 +1404,25 @@ def sort_values(self, inplace=False, ascending=True, na_position='last'):
             return self._constructor(values=codes, categories=self.categories,
                                      ordered=self.ordered, fastpath=True)
 
+    def _funcs_for_rank(self):
+        """
+        For correctly ranking ordered categorical data. See GH#15420
+
+        Ordered categorical data should be ranked on the basis of
+        codes with -1 translated to NaN, as floats, and unordered
+        as objects
+
+        Returns
+        -------
+        numpy array
+
+        """
+        if self._ordered:
+            f = _algos.rank_1d_float64
+        else:
+            f = _algos.rank_1d_object
+        return f
+
     def _values_for_rank(self):
         """
         For correctly ranking ordered categorical data. See GH#15420
@@ -1416,10 +1435,12 @@ def _values_for_rank(self):
         numpy array
 
         """
-        values = self._codes.astype('float64')
         if self._ordered:
+            values = self._codes.astype('float64')
             na_mask = (values == -1)
             values[na_mask] = np.nan
+        else:
+            values = np.array(self)
         return values
 
     def order(self, inplace=False, ascending=True, na_position='last'):